I had generally considered the fuse as protecting against wiring problems -- wires being damaged / pinched, etc., or the motors shorting (motors often fail open, but also often fail by shorting together part of the coil, resulting in a much lower resistance and potentially a fire hazard). You're right that they can protect the motors from excessive loads as well.
The temp sensor would provide you feedback, but there are limits to what it can protect against. The biggest issue is that on these brushed motors you really can't mount it anywhere that is directly connected to the heat source (coil inside the motor). You can connect it to the case but it's basically heating up from a combination of heat transmitted through the front & rear bearings, and the air movement through the case. If you get a true stall or very low-speed high-current-draw scenario, the case will heat up many seconds after
the winding assembly has burned up. Where it is really helpful is for extended use -- like the scenario that burned up one of my Banebots motors, even with a 40A limited step-down converter, which was a full (125+) lb load, run for an extended period of time.
In that type of situation a temperature sensor on the case is perfect and would be able to shut down the motor before the internals got too hot.
You *could* use the microcontroller to monitor current
and shut down under certain circumstances. It could even monitor current and attempt to approximate the temperature inside the motor. Power lost in the windings is just I^2 * motor resistance - if you assume half of the current goes to each motor on average, then it becomes fairly easy to estimate the wasted power in each motor. You can approximate the mass of the armature and the heat capacity, though you'll have a harder time estimating the rate of dissipation. I tried to build some amount of that modeling into my Power Wheels Motor/Gearing calculator XLS
. I had done it two ways -- one where I calculated the theoretical net power (based on the motor specs - torque & RPM) and subtracted that from the delivered power (current * voltage after wiring losses). That was a bust, because most of the motor specs are not that accurate. I moved to the I^2*R method (plus bearing losses -- that will be harder to calculate in the system without an encoder).
Otherwise, for overload protection, I'd recommend a thermal breaker (slower to trip than fuses), or continue to use a fuse. I don't think stepping up to a 40A fuse is a problem.
For an interesting visual picture of how quickly these types of motors fail when fully-stalled or under heavy load, take a look at the "Locked Rotor Stall Tests" on Vex Robotics' page, eg this one for the Banebots 18V motor (run at only 12V):
Even at just 8V, stalled, the motor fails after about 12 seconds (you have to download the CSV file to really see it - the current decreases from ~66A to 47A as the motor heats up, then suddenly jumps around 13 seconds when the wiring shorts out, then goes to 0 at 21 seconds when the short finally blows out).
The 'Banebots 550' motor is the most similar to the stock Power Wheels ones, and it lasts just 3 seconds at 12V. (pulling ~70A). They do their testing with larger-capacity wire and a strong power supply -- the typical SLA batteries will sag to 7-8V if you try to pull 80A out of them, so the "stall" rating on the motor is never actually seen in ride-on cars. At '30-35A' (6V), they go about 30seconds before smoking -- but that's from a "cold" starting point.
These little fan-cooled motors pack a lot of power into a small size, but they just don't have much thermal mass. Ccompare that to the CIM motor listed on their page,which is a sealed larger motor with ~350W of output -- basically like 2x the stock Power Wheels motors. It can handle 12V locked for ~37seconds before failing, and that's pulling 80-90A for most of that duration.
F-150 - 24-36V with homebrew 100A variable-speed motor controller, 4x 775 motors (no, it's NOT all-wheel-drive--look HERE
for more details!)
Pink Rocket Princess Mustang - 15mph, 36V, 4x 775 motors, homebrew PWM controller rev 2....
Hurricane - 24V w/50A step-down controller, 775 motors